Beyond Isolated Words: Diffusion Brush for Handwritten Text-Line Generation

Gang Dai, Yifan Zhang, Yutao Qin, Qiangya Guo, Shuangping Huang, Shuicheng YAN

Keywords: Handwritten Text-line Generation;Image Generation;Diffusion Model

Abstract: Existing handwritten text generation methods typically focus on isolated words. However, realistic handwritten texts require attention not only to individual words but also to the relationships between them, such as vertical alignment and horizontal spacing. Therefore, generating entire text line is a more promising task. However, this task poses significant challenges, such as accurately capturing complex style patterns including both intra-word and inter-word patterns, and maintaining content structure across numerous characters. To address these challenges, inspired by human writing priors, we focus on both the vertical style (\emph{e.g.}, word alignment) and horizontal style (\emph{e.g.}, word spacing and letter connections) of individual writing samples. Additionally, we decompose text-line content preservation across numerous characters into global context supervision between characters and local supervision of individual character structures. In light of this, we propose DiffBrush, a new diffusion model for text-line generation. DiffBrush employs two complementary proxy objectives to handle vertical and horizontal writing styles, and introduces two-level discriminators to provide content supervision at both the text-line and word levels. Extensive experiments show that DiffBrush excels in generating high-quality text-lines, particularly in style reproduction and content preservation. Our source code will be made publicly available.

Primary Area: generative models

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.

Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 2361